Biometrical Letters Vol. 50(2), 2013, pp. 117-126
|
DETECTION OF OUTLYING OBSERVATIONS USING THE AKAIKE INFORMATION CRITERION
Andrzej Kornacki Department of Applied Mathematics and Computer Science, University of Life Sciences in Lublin, Akademicka 13, 20-950 Lublin, Poland, e-mail: andrzej.kornacki@up.lublin.pl |
For the detection of outliers (observations which are seemingly different from the others) the method of testing hypotheses is most often used. This approach, however, depends on the level of significance adopted by the investigator. Moreover, it can lead to the undesirable effect of "masking" of the outliers. This paper presents an alternative method of outlier detection based on the Akaike information criterion. The theory presented is applied to analysis of the results of beet leaf mass determination.
outliers, entropy, Akaike information criterion, Dixon test, Grubbs test
